TY - JOUR
AU - Wakeman,, I.
AB - Abstract Dynamic pages and the increasing number of functions that websites can perform are changing users' relations with the Web. Little has been reported on how the experience of using this kind of interactive site differs from the ‘point and click’ interactivity of the early Web. This paper reports on a qualitative study of users who entered text while visiting a website of their choice. It shows how the process brought with it two levels of awareness: that of the interface, and that of the social context beyond the interface. The paper goes on to describe the perception of audience that emerged from analysing users' accounts. It also gives details of the data collection method, which is based on the work of Vermersch and has not been widely used outside France for analysing interaction with computers. The implications for website design are considered. 1 Introduction The Web is no longer a series of static pages linked together. Users are now offered many functions through sites, such as booking, shopping, calculating, customising or chatting. All of these anticipate activity from both the user and the site. Users must be able to specify their needs; the site should produce the desired response. So most interfaces that support activity offer an entry device to collect information. These appear in an assortment of guises depending on function: there are search engines, registration pages, shopping trolleys from which to purchase goods, booking forms, query facilities, comment boxes and bidding solicitation for online auctions. Most of these devices specify how and what information is to be entered so that it can be processed automatically, or at least consistently. But, by and large, they require users to type in the details. These devices are clearly interactive — in that the behaviour of users affects the response they receive — but this is not the ‘point and click’ interactivity of following links. These devices do more than provide the freedom to wander through a hypertext structure at the user's own pace. So is there a change in the quality of interaction going on? And, if so, which constructs might be appropriate for describing and exploring users' behaviour in these new contexts? This paper puts forward evidence that users go about their business on websites with two levels of awareness: that of the interface, and when users become involved in entering text, that of the social context beyond the interface as well. In entering text as part of interacting with a site, users start to perceive their behaviour in terms of person-to-person rather than person-to-machine relationships, even though they may know that they are addressing software. The evidence for this assertion is drawn from an analysis of users' language choices as they describe their behaviour and thoughts after using a site and entering text into it. Emotional responses, speculations about ‘audience’ and references to self presentation are also touched on in this context. The paper concludes with a discussion of the implications for design that come out of this analysis. The interviews in this study were conducted using Vermersh's ‘explicitation’ technique (1994) for collecting accounts of thought and behaviour. It has not been used extensively outside France, or for work with computers. This restricted dissemination is reflected in an enlarged methodology section and a short discussion at the end of the paper. 2 Theoretical background Human–computer interaction (HCI) has a history of looking at the relationship between human and machine in cognitive terms; this has led to an emphasis on problem solving as an analogy for how machines are used by individuals. By contrast, the research presented here focuses on the Web as a medium rather than a tool and thus we employ a social framework for understanding this engagement. Payne (1990) uses Clark and Schaefer's theory of conversational contributions to attempt an analysis of interaction. Clark and Schaefer suggest that participants communicate and, at the same time, attempt to ground this in feedback routines that demonstrate the success of their mutual understanding (1987, 1989 in Payne 1990). Payne applies this conversation model metaphorically to working with a computer; a system that acts and at the same time must indicate that is acting and how, seeking ‘acceptance’ at each stage of a task. Applying this framework he is able to show that these interactions have similar rules to the turn taking of spoken engagement and that, when these are violated, users encounter problems with understanding the behaviour of the machine. Reeves and Nass (1996)’ book ‘The Media Equation’ finds ‘that individuals’ interactions with computers, television and new media are fundamentally social and natural, just like interactions in real life’ (p. 5). It is worth noting here — since this paper is only concerned with the Web — that they do not give a definition of ‘new media’. ‘Media’ is taken ‘in its broadest sense, including communication technologies’ (p. 259), which might, thus, include non-representational media like telegraph and telephone. Despite this, they neither consider networked media as a separate case; nor discuss situations where there may be genuine communication, perhaps mediated by computer. However, this looseness in defining scope and applicability does not weaken the general thrust of their findings for interactions involving non-networked computers. They conclude that, for want of a better approach, people fall back on social interaction skills and expectations when dealing with representations of people on television and, equally, when working with computer interfaces. More specifically, they find that people respond in distinct ways to machines given different ‘personalities’ by different interaction styles and that people distinguish different voices used by the same machine as different social actors. While spoken language is not under consideration in this study, and matters of what might constitute ‘a personality’ have been left for others to debate, the issue of using language to create identity is central to the discussion here and we will return to these findings later. Within a cultural studies framework, Morse (1998) has explored the notion of the subject in virtual environments, suggesting that television prepared us for machine subjects that address the viewer directly ‘by pretending to talk to you’ (p. 8). She looks at the shift in pronouns as ‘machines will come to employ “I” and “you” with greater ease, speaking in personal modes of address’ (p. 9). These studies seek to discuss interaction with machines by relating it to interpersonal communication. The study reported here follows this tradition, being situated within a framework based on human interaction derived from Goffman's work on social cues. Goffman (1981, p. 18) theorised on the nature of human interaction in terms of social rules, which he called ‘ritual constraints’, where ‘ritual is concerned with the expressive implication of acts, with the sense in which acts may be read as portraying the position the actor takes up regarding matters of social import — himself, others present, collectivities’ (p. 37). This is a development of his work on the deliberate and instinctive acts of self presentation that accompany social encounters (Goffman, 1959, 1969) as parties signal their intimacy, group allegiances and status to each other by adopting roles which present particular aspects of themselves. Goffman's analysis is seen as significant here for the light it sheds on the interviewees’ descriptions as they attempt to choose the right role to present in the face of very little information about and from the other party in their interactions. Goffman (1981) also attempted to develop a comprehensive list of the participant roles that people might adopt, or find themselves using, in any verbal exchange. His work was taken up by Levinson (1988), who expanded his model. Even this was felt to be insufficiently broad when applied to email relationships (Pemberton 1996) because new relations between participants came with the possibilities afforded by the technology. This paper does not seek to explore the success of the model for interactions through websites — although this would be an interesting project — but acknowledges a debt to the concepts and terminology of Goffman and Levinson in analysing whom users might be addressing. As Pemberton's work indicates, analysis of participant roles has been undertaken in the field of computer-mediated communication. There has also been analysis of computer transmitted messages, using varieties of discourse analysis — as is used here — and quantitative methods. Considerable work exists exploring the use of communication tools such as email (e.g. Sproull and Kiesler 1992), customised packages for discussion (Winograd 1987; Adams et al., 1999), mailing lists, newsgroups, chat sites and virtual environments (Benford and Fahlen 1993). Work charts the evolution of cultures particular to online environments, development of systems that support communication more effectively and the impact of technology on communication. Some of this research concerns communication on the Web, but the focus is usually within that subgroup of sites that specifically host discussion between participants (e.g. Erickson 1996). Even within this arena, very little work has looked at the role of the site's host, although there is, for instance, Dibbell (1994) on the effect of mediation — or lack of intervention — in a multi-user dungeon and Light and Rogers (1999) on user–producer relationships in Web news forums. Work that is more closely related to the study reported here has now started to appear in commercial contexts as companies and their consultants explore people's attitudes to ecommerce and spending money online (e.g. Cheskin 1999). The focus of these user studies has been upon issues of privacy and security on the Web; issues touched upon here in a wider context. This study — in that it deals with users' experiences of general-purpose websites — is in an area where traditionally analysis has been in terms of use and usability at interface level. There have been studies (e.g. Graham-Cumming 1997) of how users move through webpages, some drawn from observation; some from analysis of site request logs; some using eye-tracking devices; as well as many surveys and interviews. Key issues have concerned with how to manage information — searching for it, structuring it, producing navigational aids for it — not relationships. Work on text entry during form filling — largely paper-based — also primarily concerns itself with interface issues like clarity of design (e.g. Wright 1984). While this paper is concerned in part with interface design, it also seeks to challenge the dominant conceptualisation of the Web as a space, rather than a series of connections. Navigation in hypertext environments has been discussed at length (e.g. Nielsen, 1990; McAleese, 1993). It is the interactivity of linked pages in structures, which support information retrieval. For instance, there is ‘up’ and ‘down’ to different tiers of information in a hierarchical structure. In a new environment, such metaphors helped ‘anchor’ the user and make information manageable, to lesson the risk of getting ‘lost in hyperspace’ (Hammond and Allinson, 1989; Edwards and Hardman, 1993). As a consequence of this dominant information structuring strategy, Web browsers — chiefly the big two: ‘navigator’ and ‘explorer’ — were designed to support a geographic concept of the Web. By grouping files as ‘sites’, users have been encouraged to wander around places constructed and put at a distance by producers. Now that website producers — and, by proxy, their software — can make rapid changes to the files that users access, or create new ones on the fly in response to user behaviour, this remoteness is no longer as appropriate as it was when a finite system sat on the user's hard drive. The access mechanism to files is interactive: a protocol that transmits requests across the Internet and responds with the required material for presentation as multimedia by the browser. At this level the language associated with the Web is drawn from conversation terminology. The perpetuation of a space metaphor hides this fundamental interactive mechanism and puts the originator at a distance that is no longer as valid in a networked environment, especially a dynamic one. A challenge to this model is needed. As the range of functions offered by the Web increases, we want to be exploring design language for the interface using terms and metaphors drawn from communication. 3 Methodology In approaching this study, we wanted to know: What thoughts went through the mind of the users as they approached and started the task of entering text into websites? How did these thoughts compare with their thoughts when using other parts of the site? Was there justification for using a communication metaphor to describe interactions with the site? Thoughts are, of course, impossible to collect in unmediated form. However, there is a history of collecting verbal accounts from people — by questioning them, recording them or working from written records — with a view to understanding more about their activities and beliefs. Within HCI, accounts of computer-related activities are analysed for what they reveal about the activity's components or the difference between users' actions and explanations when testing a system (e.g. Dix et al., 1993, pp. 242–243; Norman, 1988). In other social sciences, the activity under review may be the giving of the account itself. Discourse analysis has focused on how language is used in accounts to construct versions of the social world and how its use varies with the functions that people are trying to perform with it (e.g. Potter and Wetherell, 1987). In this study, we construed the thinking that we wanted to investigate as a series of mental activities stimulated by — but not necessarily directly related to — the experiences users had with the sites in front of them. We decided to collect users' accounts of the thinking they did — closely associated or not with the task of using the website — during the period of conducting the task, for analysis as constructions showing users' interpretations of the Web. This seemed the closest we could come to actually reviewing the thinking involved, but it raised a number of methodological questions: What exactly should the accounts be based upon? When should the accounts be collected? How should the accounts be collected to gather data at the required granularity? 3.1 The basis of the accounts In pilot studies, we compared using various experimental conditions — which had the advantage of producing results with controlled variables for straightforward comparison — with giving users freedom to describe an experience that they had entered into through personal motivation. It quickly became apparent that if we controlled the conditions of the experience we could affect the content of the account we received dramatically, particularly in the area of motivation. Since motivation in conducting an activity is usually a controlling factor in the perception of the activity, this immediately reduced the study's scope and credibility. 3.2 Collecting the accounts It is possible to collect accounts of an activity at different times and in different conditions: during the activity or after it; in front of a recording of the activity, in the same place that the activity is usually performed or somewhere unknown to the interviewee. Unsurprisingly, these factors were found to affect the accounts. Describing concurrent verbal protocol, Ericsson and Simon (1993, 2nd ed.) discuss the set of conditions in which verbalising during a task affects ‘the course or structure of the thought processes’ (p. 106). They conclude that telling the subject how to verbalise can change thought processes by calling for certain information at the expense of that the user ‘would normally have attended to’ (p. 107). Ironically, in requesting all thoughts, even trivial associations that might not normally be volunteered by users (in that their perceived relevance is low), the study entered the condition of asking the users to redirect their focus. The result was either no account of incidental details, or elaboration that took so long that normal thought processes were interrupted. This was an unsatisfactory state of affairs, but, although concurrency was a problem in this situation, an obvious drawback with retrospective methods was that of accurate and detailed recall. Various approaches could have been taken to address this. Prompting the memory by placing the person giving the account in front of a recording of their activity has been used to focus the interviewee on the chronology and details of the activity. However, we had defined the activities we were investigating as mental, at one remove from the tasks the user conducted to explore the site and enter text. Our finding was that new thoughts occurred to the users, generated by the experience of watching themselves and the ground they had covered in conducting the Web task. These new thoughts interfered with remembering details of the mental activities taking place at the time of the recording. This method also posed the challenge of recording users conducting their Web task ‘naturally’, to avoid affecting the motivation with which they entered into the task. Prompting can be entrusted entirely to the interviewer in the shape of questioning. Though this can easily result in an account of thoughts led by the interviewer and tailored for the environment and interaction of the interview, no method is free from the potential for this distortion. In the end, it was decided that the integrity of the original experience was paramount. We therefore chose to explore methods of retrospective questioning. Applying Vermersch's explicitation interviewing technique overcame a few of the methodological problems described above, though not all. It was able to produce accounts that offered the right degree of granularity, though these accounts may have included features associated with all retrospective explanations, such as post-hoc rationalisation and simple forgetfulness. However, the technique takes on board both of these issues. It relies on a specific style of interviewing, which avoids inviting judgement and uses evocation to assist with recall. It is specifically non-directive in its questioning to avoid leading the interviewee. It has been used for several years to evoke cognitive processes retrospectively, particularly in the French educational system, with pupils that have failed to grasp abstract concepts. This method of data collection may be unfamiliar to readers. It is introduced here as a complementary strategy to add to the HCI researcher's repertoire. The next subsection explains the technique in more detail. 3.3 Vermersch's explicitation interviewing technique This method, through questioning, encourages interviewees to enter into ‘a state of evocation’ and hence to recall action, thought and feeling during a specific instance of carrying out a task. The method provides a detailed account of actions linked to responses and associations, at a level that includes much information usually dismissed by interviewee and interviewer alike as irrelevant. A contract is agreed between interviewer and interviewee about the nature of the interview. Then the interviewee is encouraged to think of a particular episode and go into a state of evocation so that it can be described in detail. If the episode is part of a series of similar events, then one — the first, last or the most memorable — should be fastened upon. The interviewee is helped into the evocative condition by sensorial questioning: A/RG Just put yourself back into the situation. Don't tell me a story, just put yourself back into the situation and tell me exactly what you did. Was it morning or afternoon? It would have been afternoon. And where were you? I was in the Lab. It was at that terminal there. And was it a hot afternoon? Was it a cold afternoon? Um, not so I noticed either way. Certain cues, such as the unfocused gaze of the interviewee and their gestures reveal whether they are in evocation or not. Sometimes there will be moves in and out of evocation in an interview, but the purpose is to foster an environment where evocation is dominant. The interviewee is steered away from any generalisations and comments, such as: ‘Whenever I…’. If they offer an opinion, it must be clarified whether they thought it at the time or are telling it to you as part of an explanation now — if the latter, it is politely dismissed. The intention is to get an account that, usually chronologically, describes the event as if the interviewee were conducting it again, rather than an account designed for the listener. Even with the interviewee in a state of evocation, questioning is necessary for guiding the interview. To get a high level of detail, or just to maintain a flow, prompting is needed. Prompting can take many forms, from echoing, to specifying: ‘When you say you did X, what did you do?’ to clarifying ‘I want to understand. Was it like this, or this?…’. It does not take the form of a closed or leading question. The interviewer should not introduce their own presuppositions about the possible form or content: for instance: ‘what did you see, or hear, or think, or whatever?’ rather than ‘what did you see?’, bearing in mind the huge interpersonal differences in mental processes. If a reason for an answer is sought, then careful questioning using ‘how’ and ‘what’ can be made to cover the same ground as ‘why’ questions, as ‘why’ tends to force judgement and retrospective rationalisation: ‘How did you know that X?’ ‘What were you thinking at the moment when X?’. (for a fuller description of the technique, see Vermersch 1994 — in French; Vermersch 1999; Light 1999). 3.4 Design of study Twenty Web users were interviewed. The interviews were conducted near, but not at, the user's machine in what was a familiar environment. All the interviews were one-to-one and the interviewer attempted to maintain a consistent manner to all interviewees, quietly encouraging, but interrupting politely to refocus or clarify. Interviewees were asked to describe as fully as possible the last occasion upon which they had visited a website and, through the course of their visit, entered any text. They were encouraged to go into meticulous detail, to relive the experience as it happened rather than trying to be relevant to the interview situation in any way. This kind of instruction can never be wholly effective, but some interviewees exclaimed after the interview at how much they had remembered that they did not know they had thought, which may be regarded as an anecdotal measure of success. Although interviewees knew — through selection — that the entering of text was of interest, this knowledge was not considered prejudicial to the aspects of the accounts under investigation. However, visits to general purpose ‘search engines’, such as Yahoo, Excite, or AltaVista, were excluded from the qualification criteria. These sites have a dominant identity as search mechanisms, meaning that entering terms is the primary motivation for visiting these sites, and this was deemed outside the remit of the study. Interviews lasted on average just more than half an hour and dealt with between a minute and 10 min of behaviour and thoughts. They were recorded in audio only. This provided 12 h of material for transcription and analysis. Analysis was concerned with variations within accounts and patterns between accounts. We looked for signs of relationships. These did not have to be straightforward statements from interviewees. In fact, interviewees had been given no idea, which details were of interest, so that their accounts would not be prepared with our research agenda in mind Reeves and Nass (1996) note: Participants in our experiments were not aware that they equated media with real people and places. Therefore, attempts to verify the media equation can't rely solely on talking to people, listening to their stories, or asking them questions on a survey. Social and natural responses to media are not conscious. In as much as we were researching in a similar area to their ‘media equation’, we faced similar problems to Reeves and Nass. It would have offered no benefit to acquaint interviewees with the particular direction of the analysis by asking direct questions. Instead we looked for evidence of more or less conscious social behaviour and references that indicated where the interviewees peripheral awareness lay. Draper (1988) warns against limiting the search to ‘surface clues and markers’ in analyses of this kind (p. 15). Because discourse analysis was being used, another researcher — who had not been present at the interviews — was asked to examine the transcripts for signs of leading questions and, subsequently, three colleagues including the interviewer looked at the findings to draw independent conclusions. These concurred. Nonetheless, the findings are quoted extensively below so that others can make up their own minds about their significance. 3.5 Participants These were people who had recently entered text into a website. Participants were picked — through a brief semi-structured interview — who used the Web as part of their everyday life. All interviewees accepted the Web as a practical and even obvious resource, though they were not necessarily enthusiastic about it. This familiarity seemed appropriate for several reasons: to get the longer term picture, as more people become experienced and interactive behaviour more central to media; to ensure that the accounts did not dwell on teething troubles and the anxieties associated with learning to use a new technology, obscuring other findings; and to deal with a group that were not easily dismissed as naı̈ve and therefore more prone to confusion. Otherwise, we attempted to get a balance based on the ‘experienced’ Web user demographics of age and education as defined in the latest GVU survey (1998). The youngest interviewee was 21 and the eldest, 53 years old. Most had a first degree, but the occupation of interviewees ranged from lighting designer, to personnel officer, to researcher in electronic engineering, while a three were still studying. Nine participants out of 20 were female; this does not reflect current Web demographics, but, unlike in other sectors, the ratio is changing fast (Computer Economics 1999) so a judgement was made to increase the percentage from 27.5% (GVU 1998) to 45% in this sample. The sample deviated in one other key aspect: all the interviewees were European English speakers, whereas the largest number of users — and many surveys — are based in the US. 3.6 Participant roles used in the analysis As mentioned above, the idea of participant roles comes from Goffman via Levinson (1988). Since everyone involved in the study described a situation where they individually composed and entered their text in one sitting, there is no need for an elaborate framework on the author side — speaker and information source are synonymous. On the reception side: ‘producer’ is being used as a blanket term for references to the body of people represented by the website: company, company staff, content contributor, designer or individual developer. In this respect the use differs from Levinson (1988), but conforms to the convention of juxtaposing it with the term ‘user’. ‘audience’ refers to everybody that has access to information entered by the author. ‘perceived audience’ is everybody the author believes has access. Within the audience may be found: ‘ratified recipients’ such as company staff; ‘unratified recipients’ such as ‘eavesdroppers’ in the system, or people to whom the ratified recipients have passed information, for instance, by selling on email addresses; and ‘target recipient’, the person, team or software that the author intends to receive and act on the information. 3.7 Transcription notation The transcription from audio to text lost paralinguistic data on the way. To correct for this as far as possible, punctuation was inserted to give a sense of the utterances, while some crude indications of behaviour were included, such as references to laughter. It is this version that has been used for quoting here. Where sections of text have been excluded from the sections quoted, the symbol {…} has been inserted. Interviewer comments and questions appear in italics. All quotes from transcription are labelled with the segment (see below) and interviewee initials. 4 Findings This section starts with a brief overview of the context of the users' activities, their motivation and the effect of interaction on achieving their goals. The main findings follow. To explore these central aspects, the accounts were divided into segments relating to the kind of explanation being given. Six kinds of segment were identified: explanation of introduction to the site, explanation of the user's purpose, explanation of moving through the site, explanation of thoughts on moving through the site, explanation of entering text, explanation of thoughts on entering text. The segments did not necessarily appear in this order in all accounts, and not all interviewees dwelt on A and B. However, segmenting in this way allowed comparison within and between accounts. In the descriptions below, where part of an account is quoted, the letter preceding it reflects the kind of segment it represents in the account from which it was drawn. 4.1 Overview The table gives an overview of the activities the interviewees described and contextual information extracted from their accounts to be discussed below. ID . Activity . Entry mode . Content . Branding . Target recipient . BL,CB,CS,JD,MI,MM (Register to) explore site Custom form: text Identity details Media JL,LB, RLSJ Book holiday, conference or flight Custom form: text, menus Financial, identity and product details Service company/organisation Self and (company staff/system) JF Subscribe to hard-copy magazine Custom form: text, menus Identity details Media (subscription system/staff) AC Subscribe to online service Custom form: text Financial and identity details Media (subscription system/staff) IH Connect to email service Custom form: text Identity details Service company (subscription system/staff) PG Enter student data into database Custom form: text Information relating to others Customised management tool self and university administrators BL,CB,CS,GP,MI Search for information Field: text Term(s) relating to purpose Media/company/homepage Search software TO (Register to) download software Custom form: text, buttons Identity details Manufacturing company RG Buy software Custom form: text Financial and identity details Manufacturing company Company system/staff MM Assess value of user's car Custom form: text, menus Information relating to car Media Search software AH Register user's website with search engines Custom form: text Identity details and information relating to site Service company Registration software and staff at Yahoo SB Explore site (and leave comment in guestbook) Comment box: text Message Homepage Author and public IH Complain to internet service provider Email form: text Identity details and message Service company Company representative AT Order items to collect Custom form: text Identity and product details Manufacturing company Software → remote staff ID . Activity . Entry mode . Content . Branding . Target recipient . BL,CB,CS,JD,MI,MM (Register to) explore site Custom form: text Identity details Media JL,LB, RLSJ Book holiday, conference or flight Custom form: text, menus Financial, identity and product details Service company/organisation Self and (company staff/system) JF Subscribe to hard-copy magazine Custom form: text, menus Identity details Media (subscription system/staff) AC Subscribe to online service Custom form: text Financial and identity details Media (subscription system/staff) IH Connect to email service Custom form: text Identity details Service company (subscription system/staff) PG Enter student data into database Custom form: text Information relating to others Customised management tool self and university administrators BL,CB,CS,GP,MI Search for information Field: text Term(s) relating to purpose Media/company/homepage Search software TO (Register to) download software Custom form: text, buttons Identity details Manufacturing company RG Buy software Custom form: text Financial and identity details Manufacturing company Company system/staff MM Assess value of user's car Custom form: text, menus Information relating to car Media Search software AH Register user's website with search engines Custom form: text Identity details and information relating to site Service company Registration software and staff at Yahoo SB Explore site (and leave comment in guestbook) Comment box: text Message Homepage Author and public IH Complain to internet service provider Email form: text Identity details and message Service company Company representative AT Order items to collect Custom form: text Identity and product details Manufacturing company Software → remote staff Open in new tab The following notes provide a brief clarification of the terms used in the table. In each column the descriptions given have been drawn from the accounts, but simplified for the purposes of grouping: ID: each pair of initials corresponds with one interviewee. Some interviewees described more than one text-entering task in their account. Activity: what the user intended to do in visiting the site (with the text-entering task in parenthesis if it had no relation to the user's original intention). Entry mode: the kind of interface device and what entry modes it accepted. Content: what kinds of information the user entered. Branding: how the site's producer was primarily identified- ◦ Media — seen in terms of a related print or broadcast medium ◦ Service company — seen in terms of a service supplier ◦ Manufacturing company — seen in terms of a product supplier ◦ Organisation — seen in terms of a body of members ◦ Homepage — seen as the work of an individual Target recipient: this is stated where interviewees were sure of who or what would receive (and might interact with) the information sent, placed in parentheses where only a hazy or contradictory impression was given, or left blank. 4.2 Motivation All the interviewees brought a personal — social or business — goal to visiting the site they described. This was not a prerequisite; they might have described entering text on a site that they had found serendipitously. However, in each case, the site or, once, the kind of site — ‘Um, I actually did a search for D…. because I particularly liked the books.’ (SB) — was chosen as part of performing the activity. Only in the case of the subscriptions was completing a form considered a primary reason for visiting the site, however in many cases entering and sending information was expected because a transaction that required it was involved. By contrast, commenting in the guestbook and filling in registration forms occurred because the means were provided — and in the case of registration, demanded. This latter discovery was unwelcome and became the cause of deliberation; the form's presence nearly deterred users from pursuing their intentions — and had, they suggested, on other occasions: EF/TO I quite often would back out on something like that. Right, but on this occasion you didn't back out, you … No, no, because I wanted that software. {…} I don't see why there would be any other sort of obstacles at that point. You say ‘obstacles’…. Well, um, it's. I mean, the original link — which I assumed would give me the software — wasn't the link, it was a link to another page which led me on to the software eventually, so it wasn't a direct route to what I wanted, so…um, that's an obstacle. I mean it's just standing in the way of just downloading the software, which is the goal — what I'm doing. In three cases, the fulfilment of the broader goal was postponed midway through completing a form — once because the form was too inflexible to serve the user's needs and twice because of technical hitches. In all three cases the task was finished by telephone. The motivation to complete the activity was strong enough in each case to dominate interviewees’ descriptions of their behaviour and thoughts, though not to the exclusion of all other observations. The next two subsections explore the implications of these other observations and how they were expressed. 4.3 Awareness of the interface Interviewees’ accounts show passing consideration of interface issues as well as the pursuit of their goals. The accounts refer to the appearance, functionality and the ease of use of the sites. For instance: C/AC Yes. Ok. Uh, as I recall there was a big blob of colour in the middle. Uh, I cannot remember what was underneath, but the pointer changed to a hand, yeah, and so I didn't bother reading the rest at the bottom. and: C/CS It's to do with the big advantage of paper text over computer material, {…} something about on the G… adverts that you know by the typeface pretty much whether you are interested in them or not, you know what, say, a public sector job — which is what I was vaguely thinking I would see if there were any of — what a public sector computing job is going to look like. You know what it's going to look like, so I was kind of expecting to scan these jobs and find them. {…} what happened was, as I started to use the site I realised that it wasn't going to give me that at all. Moving through the sites brought up references to links and allusions to space. For instance, there were specific uses of the ‘navigation’ metaphor: C/JF …lots of very, very nice graphics, but fell over a couple of times when I tried to navigate through the site proper. C/LB It seemed quite, well, there was no problem navigating the site anyway — it was quite easy to get the information you wanted. C/AH The site was in frames: the left-hand column frame was navigation throughout the website… Some users encountered menu selection options (see table). Accounts of using these devices range from the uneventful to the furious. A user entering address details: E/IH then it says ‘county’ and you are not allowed to fill in the county, you are given a choice…which doesn't have East Sussex or West Sussex, it just has Sussex, which (laughs) {…} Um, then it says ‘address’: ‘street’, ‘number’, put that in, ‘town’, put that in, ‘county’ and it gives you this choice… and, slightly annoyed, I picked out Sussex. and another entering dates for travel: E/JL it was relatively easy to fill in except for with drop-down menus. You've got the whole thing, you know, you click the little arrow next to the, it comes up saying something like the 14th and you click the little arrow and you are always completely in the wrong place for the date that you want, so you've got to kind of like scroll up and scroll down which is just really annoying and in a way, you would, well, I would almost rather do it freehand than that. Same with the month, it always comes up as June or something, so every time you're kind of trying to test out a combination you have to kind of scroll it, or whatever, to the month… 4.4 Awareness of others Given the number of observations about the interface, it is not surprising that many of the accounts show use of the pronoun ‘it’ for describing features of the site they visited (see E/IH above, where the use of ‘it’ to refer to the site has been placed in italics). However, all the interviewees switched between using the pronouns ‘it’ and ‘they’ in describing features and they did so in a discernible pattern. Both terms could have been used in the situations to which they were applied, with only a modest difference in the superficial sense of the utterance: E/LB …because, you know, they've got to — what do you call it? — check your, this was a debit card so it has to actually go and check whether your funds are available, validate your card. E/MM It was basically just a, a form and I could see that they wanted your email address, um, your telephone number and your home address. While there was a degree of interchangeability between actual use of the terms, it was possible to identify where the use was likely to change from ‘it’ to a mixture of ‘it’ and ‘they’. Some factors affected this tendency, but accepting these, there was a pattern to the usage. When interviewees were giving an explanation of their thoughts on entering text (F) everyone was using ‘they’ from time to time, if not all the time, whereas only two people used ‘they’ in segments of the kind A, B or C. This pattern was considered significant, because the use of ‘it’ suggests engagement at interface level with the information and structure provided, whereas the use of ‘they’ implies an awareness of other people and therefore a different kind of relationship with the site. In the explanation of their introduction to the site, a couple of users started their account by referring to the company identified with the site, especially if they had found the site in order to purchase a product from the company. In one instance, where the enthusiasm for the producer was high, the interviewee adopted the use of ‘they’ interchangeably with it’ to refer to site features throughout his account. B/RG It basically came down to the fact that I wanted a personal copy of Red Hat — the next latest version — um, so went.. Red Hat's got a site, so I went there. It's actually a very good site with lots of information and all sorts of stuff on it. Just cool. Looked up… they had, they had good links for the, the bits I was interested in, so I followed them. Interestingly, this user reverted to the pronoun ‘it’ when describing the site that Red Hat referred him to in order to monitor the progress of his parcel. One user who described a site that was presented as an individual's homepage showed a similar tendency: C/SB Um, what I first saw was actually quite a quite a decorative site and he'd put a very sort of bright blue background with sort of, not silhouetted, but, um, sort of the shapes of dragons in the blue so they weren't sort of pictures of dragons but you had the shapes sort of as a suggestion in a slightly lighter blue on the background. Um, he had sort of the usual “Welcome to Avatar D….. Site”. Um, he had a quick brief bit about himself, basically saying, if I remember rightly, he said he was a computer programmer from somewhere in America. In this case the producer of the site has introduced himself up front and the user refers to ‘he’ and ‘him’ on most occasions when describing the site in her account. Other sites do not automatically get this treatment from her: E/SB He hadn't done much to sort of make the book itself. I know some that I've been to, when you click on the button, it has a big picture of the book at the top and then an entry field. His was just a plain entry field and underneath you could view the guest book if you want. Another user, describing her experience with a page presented as a homepage (it was labelled ‘X's page’), actually observed her tendency — which was less pronounced than the previous user's — to talk about the producer when describing the page: C/GP I thought that it was edited by someone, it's a homepage, it's a bizarre thing — it's a resource and it could be a homepage, but there's no personal information about her, which is bizarre. I mean, I haven't looked, to be honest, but there is no obvious link to personal information. {…} it's structured, they stress the point that it is highly edited and what you've got there is quality stuff because somebody has spent lots of time on the net trying to find the… so maybe they, ‘they’ again, you see, it's bizarre why I use this kind of terminology, the personalised terminology about this page — I wouldn't use it for something else maybe… These two conditions have in common that the user is considering the producer from the start of their experiences on the site. In the first case, the user comes in with a strong sense of the company he wants to buy from, expressed as a series of very positive statements: C/RG So I typed in the Red Hat URL directly from memory. So what did I see? Saw the page. The page had a nice big Red Hat logo as their major picture at the top…um, appealing to the eye. This illustrates an effective use of branding in the conventional sense. In the second case (E/SB, C/GP), the producer raises the issue of his or her identity as an individual, especially in the first example when some biographical material is included as an introduction. This can be seen as influential branding too. Elsewhere, the dominant way to refer to the elements of the site being described was as ‘it’. ‘It’ was what the users saw, liked, didn't like, used, took issue with, and so on until they started to move into a different relationship with the site. ‘They’ appeared in interviewees’ descriptions of their response to the request for data to be entered and also appeared when the interviewee was describing how information was requested, especially if the interviewee was quoting the words on the device: E/JL They say that the airlines, it actually says that — there was quite a lot of blurb about the security of the site and I remember specifically that it said that we… ‘it is a requirement of the airlines that we have your credit card details in order to reserve a flight’, cos otherwise I wouldn't have because… I wouldn't … I mean, um, I still not incredibly comfortable about sending my credit card details over the internet… and: E/GP …in Yahoo, or AltaVista, they say. They always say how to do a search and there are differences {…} I'm quite familiar with searches and using this kind of thing, but I would always like to say ‘Teach me, tell me’. and: E/MM Oh, there's, um, at the bottom, they ask you to put in a, um, password so I think of a password. Accounts were examined to see whether it was merely an order effect: perhaps as descriptions progressed, awareness of those involved in constructing the site or handling the data developed. However, the use of ‘they’ came and went with the sections identified, whether these sections came before or after other explanations. Therefore, it may be attributed to something that took place at the time of visiting, not the time of recounting. The change identifies a move from engagement with interface issues such as content and design to engagement with the idea of others, beyond the interface. 4.5 Perceptions of ‘they’ Clearly, the use of ‘they’ has a significance in these accounts, so does textual analysis reveal more about who ‘they’ are? Two users described how they had a fleeting image of a team of designers at the other end having designed the software that the users were now engaged with. Using a search engine to explore a database: E/BL Any images come to mind? … Kind of designers, designers, a group, I don't know why. It's more a sense of people having designed that… Yeh, I had, no, I had. I had more this impression of bizarre, this stuff, it's not well done: this box comes too late. And then something like, how would they, the designers, how would they manage these keywords at the end? — because I could have put anything in the keywords. Then a kind of curiosity about what could happen next. What will they do with this information? How have they programmed… how will it work? If the word ‘they’ was echoed by the interviewer, other interviewees identified ‘they’ as the company, representative or person they believed had produced the website. However, there is a suggestion in the account above (E/BL) that ‘they’ are also in charge of receiving and handling the present information: ‘What will they do with this information?’ and this ambiguity about who ‘they’ are appears in other accounts. Personal recipients cropped up elsewhere. Here, in describing a company's policy on payment: E/AC …you fill in your credit card details and if you don't do anything then they take the first payment at the end of 30 days. If you stop it beforehand then they don't charge your credit card anything. {…} So you know I, I, I think I do it with great reluctance {…} But I guess I'm reassured because W… is the sort of person you complain to about, you know, they're sort of consumer champions so there's an element of trust. and, entering details to enquire about a holiday: E/SJ It seems a little bit more personal; it's less easy to send information like that, whereas when you're just clicking on a sign, you think it's just the next page coming up and the next image: it kind of doesn't impinge on your personal space. Once you start filling in forms, or something that, you can see a person at the other end if you like, it just feels different. It's not quite the same as just arbitrarily clicking on to— Interviewees were not asked to explain explicitly who ‘they’ are. The accounts show no distinction between ‘they’ as producer, recipient or other audient. 4.6 Social behaviour Only two of the interviewees described entering text that showed the intention of communicating specifically with other people, that is, to transmit a message of their choosing to a human audience, rather than entering specified fields or search terms that required no more than processing. The first was a complaint, entered into an email form directed at a helpdesk. The second was the only example of a publicly available, or published, communication: leaving a message on a website in a ‘guestbook’. Both writers spent time carefully presenting themselves and their message: F/SB I was very aware that what I was entering was something that was then going to be on view to other people. That it wasn't just a sort of private email that's gone to one person. It was something that would be there for other people to look at. And, even though they might not know me personally and probably have no chance of ever running into me, I wanted it to be as clearly stated and, you know, as nice as possible, sort of. I didn't want it to sound too dry or too bland. Um, but at the same time I didn't want it to sound too stupid. I wanted it to sound vaguely interesting. I suppose. What is more interesting in this context is the extent to which signs of a similar concern about self presentation affected others, involved, for instance, in filling in the occupation section in registration forms: E/JF I also remember thinking that, I think I put myself down as ‘scientist’, and I remember thinking ‘Well, what am I? Should I be put down as ‘student’ or what would I like to be known as?’ {…} I remember thinking ‘What-’, in that pathetic way, ‘how, what, how would I be categorised according to them?’ Does it matter, do I care? and: E/CB What I did was put my name in, my- and they ask you about your profession and I always wonder what is better to put in: journalist or lecturer, or other, or any. {…} but if it was a different site I'd just put down journalist or lecturer depending on which one was going to be useful. Although there may be practical reasons for choosing the best description, it is also very possible that software handles the forms, amalgamating the data on occupations into market research statistics. However, these accounts suggest that presenting the right identity choice to ‘them’ is worth a thought. These are not naı̈ve users: in the same breath one of the same interviewees talks about the destination of the information: F/CB A lot of the time I lie about gender as a matter of principle; {…} I don't feel like — I don't care if they have accurate demographics about me. (In all, three interviewees said they entered false information) So there is evidence that the interviewees in this study not only had an awareness of ‘they’ as creators, they also considered how ‘they’ would perceive them, suggesting ‘they’ as recipients. Interviewees chose to behave appropriately to a social context, even though some sounded doubtful about the value of this approach in their retrospective accounts. This occurs in the use of a range of interactive devices, and not just in using the conventional communication systems like the helpdesk form and the guestbook. Perhaps most surprisingly, this can include search engines on websites, which — on the face of it — do not appear to inspire communicative behaviour, since the interviewees using them made clear that they understood these to be software mechanisms. There is also evidence of emotional responses and of reflective responses generated by the process of entering text, which are not present in the explanations of moving through the information on sites. Already, above, issues of self presentation have been mentioned. Below, are descriptions of feelings about relationships generated by interacting. One interviewee describes hitting the button that sends his subscription form: F/AC I guess there's that moment of, sort of, no return and I guess there's a little frisson when you get to that moment which is probably quite exciting actually. Yeah. I think that's, that's why I like on-line shopping. An interviewee comparing the freedom of search engine fields to the constraints of a registration field, as she describes filling it in: E/CB If search terms don't work {…} then I think, if it doesn't work, that I should have been cleverer in the way that I thought about the search term to use, d'you see what I mean, so I look at it in a different way, so that if I put the right search term in they might come back with the right stuff and it feels much more like serendipity if it does [work], and more intelligent even though I know there's no difference, it's just that I can do more about it, {…} I don't just have to say ‘C… B…’, that's all. That's the difference. And here another example, also concerning registration: E/CS And, pressing that button [to send the registration form], how was that? That was uneventful, because I'd effectively already submitted to the process at the point when I.. do you see what I mean, I'd had to submit to registering earlier on, so I wasn't really bothered about submitting, because I'd.. I'd kind of had my emotional encounter with it. Your emotional encounter with it? Well, yeah, I didn't want to have to register. … Compare this with later in her account, when she has been able to enter the site: C/CS What you have if you look at [the newspaper] is an… array really, of a large number of jobs spread over a wide area of space that you can read very quickly, and what I got when I looked at the website was a sequential list of jobs that all looked the same. There was no chance to use visual clues to work out which ones I was interested in, and I thought ‘This isn't going to be quicker at all’. So I had a look at a few of them and then I thought that what I really needed to do was to narrow my search down… Although she is not achieving what she had hoped to as quickly as she had hoped, her response the second time is analytical rather than emotional. The explanations of how the process of entering details affected these interviewees are quite thorough and this was typical of the accounts collected. Even prompting with questions elicited very little of this kind of emotional and reflective colour during the explanations of moving through information on the site. It could be argued that moving through the information was not a primary purpose in being there — and so gets less attention — but the same can also be said of registering to use it. One attribute that seemed to inspire positive feeling wherever it was found was a confirmation of entered details: EF/JL Yeah, and you get back stuff giving you a booking reference and a telephone number and a place to ring if things aren't going well and a suggestion that you print out the page with your itinerary on it and they also email you to confirm the sale. So any more thoughts or feelings as you went through this bit? I think I was rather pleased actually. You know you kind of get it so that you think ‘Oh great, yeah… excellent!’ and: F/AC ‘Oh’, I thought, ‘this is good because this allows me to confirm, ’cause, after all, it's been three screens of information so I cannot remember, maybe, all of them, but this then compresses all the answers on to a single screen so I can quickly scan what I've given and if that's OK then it's, it's confirmed.’ {…} But I don't remember whether I felt that at the time. But, I, no, I did in a sense because I like to see that — sort of, I mean I find that careful design — that's reassuring to me, you know, that they're not sucking in everything into a void. 4.7 Other audience So far, the role of ‘they’ as recipients and the effect of this awareness have been examined. But there are also references to ‘unratified’ recipients that come up in connection with security and identity issues. Security caused concern in a way that has been recognised in numerous studies: people were reluctant to enter personal and credit card details because the Web is perceived as insecure: F/LB Ah well, slightly, ah: ‘Here's my Switch number for the whole world to listen to if they happen to be listening.’ Um, thinking ‘Well, I guess encryption works… But there was also evidence of another side to this. Not only was there concern displayed about potential eavesdroppers in the system, but also about the extent of the audience at the final destination. One interviewee talking about filling in a subscription form: F/JF ‘I remember thinking ‘Who's going to get this information’ and actually at the bottom there was a ‘terms and conditions’ text, literally in small print, I mean, um, it must have been around five or six pt in an italic font, which was completely un-, just about readable, possibly. And obviously in that situation you just go past it. Um, so it is entirely possible that that list is now being used for commercial purposes, so I'll get spammed hideously… And, here, a teacher who has entered student details into a database fleetingly considers issues of access to the entered data, and to the metadata that access logs might create: F/PG And then I had a wicked thought. I thought ‘I wonder if I could look at anybody else's.’ (laughs) {…} At one level I thought I didn't think anything, but I remember thinking, um, ‘It would be nice to just, sort of, be naughty’ — if you like — ‘and have a look at other things’ but then I thought ‘Well, they probably know who's looked at what so-’ and I just cannot be bothered thinking that someone else might know where I'd been. Life's too complicated. These accounts suggest that interviewees were aware of the possibilities for — and experienced anxiety about the extent of — hidden audience. 5 Discussion of findings The examples above show that in going about their business these interviewees were aware of two levels of interaction, one with the interface, the other with shadowy figures beyond the interface. The point at which this second awareness develops seems to be during the entering of text and the thinking associated with it, since it is not apparent in using the interactivity of navigational devices or menu selection systems. Interviewees appeared to move swiftly and easily between the two states of reference, suggesting that users might move as easily in their perceptions. There was no sense that having two different states of association affected their competence or enjoyment of the site, but it does raise some interesting questions for developers. The justification for adopting a communication metaphor is strong when dealing with interactive components on sites, though less so for describing moving through static information, where navigation terms are more helpful. The scope of each metaphor's appropriateness has not been defined, but the evidence above suggests that people use communication behaviour even when they are not sure about who or what they are interacting with. There has been little work on Web users' perceptions of text entry mechanisms and the study reported here is far from exhaustive. For instance, unlike most studies, it is based on work with European rather than American users. But some interesting findings have begun to emerge that need further investigation. And, although qualitative studies are designed to raise issues deeply rather than widely, their implications can lead directly to changes in practice. It is hoped that after examining the study here, designers will experiment with making explicit the metaphor of communication in building functionality into sites, seeing the solicitation of information as an ‘invitation’ to interact. Producers may wish to consider how they phrase and present their ‘invitations’ for maximum effect, both in terms of persuasiveness and in carving an identity. The subsections below correspond to the categories of the table at the beginning of the ‘Findings’ section and each explores the implications of the findings further in these directions. 5.1 Mode of entry Text entering requires selecting from a far greater choice of words than a menu offers. Once users are forced to think about how to describe something, they start to think about how to present it appropriately to the target recipient. If the subject being described happens to be the interviewees themselves, such as in a request for ‘identity details’, interviewees are likely to make sense of this by drawing upon self presentation behaviours that are familiar to them from person-to-person interactions. In this way, they are behaving as Reeves and Nass (1996) predict, although users in this study were in the ambiguous position of not knowing if they were talking to people or software at the other end. In ‘The Media Equation’ studies, users did not have this challenge; nonetheless, the findings seem constant across both kinds of computer use — default behaviour when asked for information by a computer is to act as if something interpersonal is taking place. If this is the case, then it follows that users' perceptions of the recipient of their information — and what the recipient's requirements may be — will affect what the users do and say next, just as happens in exchanges between people. Where users do not have a clear impression of the recipient, one might expect a less confident response to the site. There was evidence of producer choices, which became unfortunate when viewed in the context of building relationships. The use of the term ‘submit’ is one example. As indicated (CS above), the verb does not have to be seen as a direct synonym for ‘send’, but rather as arrogance in an encounter with a producer. Interviewees’ pleasure at finding the inclusion of a ‘confirm details’ stage seems to bear out that having control over one's presentation is even more than just a practical matter. Feedback — and the chance to review a contribution — reassures users that they are engaged in interaction which both parties consider significant and which both have a stake in negotiating. However, in the three accounts where a confirmation stage was described, it was greeted with surprise as well as pleasure. Before leaving this subsection, it is also worth noting the role of context in the effect of entry devices on users' behaviour. By and large, the action of ‘point and click’ is a means of selection: whether of information or a route through a document. The ‘send’ or ‘submit’ button has the same affordance (Norman 1988) of ‘point and click’ pressability as hypertext links, drop down menus, and other buttons. Its function, though, of transferring information sets it apart. While not every interviewee felt a ‘frisson’ (AC above) in using it, it represented for users a moment of no return. Consequently, many behaved differently towards it than towards buttons whose behaviour is known to be reversible in the way that Web navigation and selection decisions generally are. 5.2 Target recipient Constructions in the accounts like ‘they ask you to’ (E/MM above) show how easy it is to confound the ‘they’ that built the site with the ‘they’ that will handle the entered information. The two elements coalesce into one voice asking and, therefore, expecting to be answered. The implication from the study above is that users, left to their own devices, will expect to enter a dialogue once entering text. The kind of dialogue and the confidence with which users go into it is then a matter for development by the designers so that the voice of the site speaks clearly and makes transparent with whom the user is engaged. The websites in the study could easily be described as ‘addressing the user’ in their solicitation of particular items of information. In fact, it is possible to identify this addressee as the producer, rather than the computer interface, on the basis of the work on perceptions reported here. But interviewees showed no commensurate behaviour; they proceeded by ‘expecting’ a recipient rather than ‘addressing’ one because of the lack of information available to them. 5.3 Information content Imparting what interviewees perceived as personal information was described in more emotional and reflective terms than less personal material, although what constituted ‘personal’ varied. Although this heightened response occurred while entering sensitive information like credit card details too, the effects cannot be wholly attributable to practical issues such as information security, since, for instance, entering ‘occupation’ caused difficulties that had nothing to do with misuse of information. Occupation is a critical identifier in interactions with others, carrying associations of values, status, lifestyle and interests. If this was reflected in the interviewees’ difficulties, as it appeared it was, then it supports the notion that, on one level, a social encounter is taking place. The differences between individual users' responses can also be attributed to trust. Confidence about the recipient is referred to in connection with imparting personal information. Users who held a good impression of a company (AC), institution (RL) or individual previously (GP), had had good experiences of interacting with them elsewhere (AT) or who were told others’ good opinion of interacting with them (JL) were likely to act confidently and trust the recipient that the information was solicited for good reasons. Where they were interacting with ‘strangers’ there were more questions raised in the accounts about who wanted to know and why. 5.4 Purpose It was noted at the start of the ‘Findings’ section that for some interviewees entering information was part of achieving their purpose and for others it was a diversion. Their accounts show a weighing up of cost and gain in giving information; users completing forms with better grace if they could benefit as well as the recipient of the information, the producer. On the three occasions when people said that they had lied, each was to sabotage the information strategy of a recipient who did not seem to need what they were requesting to perform the transaction. When registration was required, hostility towards giving information was high. In each of these cases, registration was not perceived as useful to the interviewee; it was an obstacle to accessing free and unaffected products, whether pages or software. Gaining access may have pragmatically justified the completion of the form, but, logically, the absence of an additional service related to the information provided, left the interviewee talking in hostile terms about the experience. Registration can result in personalisation, but there were no examples of personalisation mechanisms used in this way in the study. These might have lessened resentment, perhaps, by offering an apparent benefit other than access. 5.5 Branding This increased awareness of the other end seems to come about, principally, as a result of being asked to enter material and planning how to respond. It is not necessarily the effect of site branding, but clearly the two interact when users start to ask about their audience. The site's branding is a clue about the identity of the other end, while the choice of language and layout also contributes to a sense of who ‘they’ might be. Working against these factors may be the clear and accurate idea that some users bring about how the site's software deals with their text, though there was not much evidence in the accounts of these thoughts at the time of actually using the devices. Thus, the perceived audience has an identity constructed from a combination of: what users bring with them in terms of expectations of the brand, users' experience with the site, especially with parts where the users' contribution is solicited, users' purpose in being there. There was a suggestion that branding the site as an individual's ‘homepage’ brought with it increased perception of the other end. With this appeared to come a certain amount of tolerance, trust and gratitude, in encountering someone who had taken the trouble to assemble some material of value, whatever its shortcomings. Companies, on the other hand, received less indulgence and benefited substantially in interviewee's eyes if they were already familiar in another context. Implicitly in interviewees’ accounts, companies started from a base of being treated mistrustfully, perceived as being unreliable or having motives of power or profit attributed to them. Companies that clearly sold something were regarded as less suspicious, since the source of benefit from their webpages was obvious. Media companies occupied a grey area — seen by some as a useful service and by other interviewees as high-handed with data, especially as data was what they were known to trade in. The interviewees tended to expect ‘for-profit’ organisations to have alternative means of exploiting websites if they gave anything away for free. It has already been pointed out that, regardless of who receives the information sent, there is a tendency to regard the text on the screen asking for input as part of the same conversation as the text sent by the user and the material returned in answer to that request. It is therefore necessary to construct one voice in handling information input, or to teach the user, through clues in the interface, about the true destination of the data. In telephone interactions with companies, assistants are usually taught to name first the company and then themselves as the representative dealing with the caller's request. This is friendlier and more accountable than acting anonymously on behalf of the company, and leads to greater confidence on the part of the caller in their interactions. This attention to detail needs to be transferred into the Web domain. It will mean recognising and handling anxieties about ‘unratified’ audience as well as sharpening focus on an appropriate target recipient so that they can be specifically addressed. It would also be interesting to consider what the engagement generated by successfully encouraging users to enter text might mean in terms of branding and, thus, in the marketing of sites. But there is no room to explore this here. 6 Recommendations This study provides evidence that a communication model is implicit in users' interactions involving text entry into websites. We have considered some of the design implications of building this communication model into sites themselves, to make the experience of using them more consistent with users' expectations. In particular, our findings suggest that the design of interactive components on websites would benefit from: an identified target recipient with a clear ‘voice’ in the dialogue, language and structure that are sensitive to the user's relationship with the producer, for instance, avoiding compulsory registration and terms like ‘Submit’, confirmation pages after the entry of complex data, with a chance to amend any feature, a simple statement about any further uses of user-provided data, positioned so users will see it, an explanation for collecting specific data that do not appear to assist in the completion of the user's task. 7 Discussion of method From the discussion above it is obvious that considerable insight was gained by collecting accounts from users in the way described. It was possible to identify patterns emerging about perceptions without asking specific questions that might have distorted the results and which, in asking about a peripheral and perhaps unconscious aspect of behaviour, might anyway not have received accurate or informative answers. This study, with its particular interest in users' motivation, attempted to combine the observations of fieldwork with the control of experimentation. It lacked some of the benefits of each method, but produced material of an appropriate content and granularity to meet the ambitious research agenda without elaborate staging or equipment. While setting out with a fixed procedure and a structured interviewing technique produces a more manageable quantity and kind of data, in this context it would have risked simplifying the relationship between language, thought and behaviour in ways that eliminate much of what was most interesting. This study is a starting point, rather than an exhaustive exploration. The material analysed here offers fuel for thinking about how websites work — or fail to work — with their users, and many avenues for further enquiry. References Adams et al., 1999 Adams, L., Toomey, L., Churchill, E., 1999. Distributed Research Teams: Meeting Asynchronously in Virtual Space. Proceedings of the 32nd Hawaii International Conference on Systems Sciences, Hawaii, January 1999. Benford and Fahlen, 1993 Benford S. Fahlen L. , A spatial model of interaction in large virtual environments de Michelis G. Simone C. Schmidt K. Proceedings of the Third European Conference on Computer-supported Cooperative Work 1993 Kluwer , Dordrecht OpenURL Placeholder Text WorldCat Cheskin, 1999 Cheskin Research and Studio Archetype/Sapient, 1999. eCommerce Trust Study January 1999, www.studioarchetype.com/cheskin/index.html as at 2/2/2000. Computer, 1999 Computer Economics, 1999. Press Release 26/5/99, Computer Economics eSite, www.computereconomics.com/new4/pr/pr990526.html as at 2/2/2000. Dibbell, 1994 Dibbell, J., 1994. A Rape in Cyberspace in Flame Wars: the Discourse of Cyberspace. In: Dery, M. (Ed.). Duke University Press, Durham, NC Dix et al., 1993 Dix A. Finlay J. Abowd G. Beale R. , Human–Computer Interaction 1993 Prentice Hall , New York Draper, 1988 Draper S.W. , What's going on in everyday explanation? Antaki C. Analysing Everyday Explanation: A Casebook of Methods 1988 Sage Publications , London OpenURL Placeholder Text WorldCat Edwards and Hardman, 1993 Edwards D.M. Hardman L. , Lost in hyperspace: cognitive mapping and navigation in a hypertext environment McAleese R. Hypertext: Theory into Practice 1993 Intellect , Oxford, UK 91 105 OpenURL Placeholder Text WorldCat Erickson, 1996 Erickson, T., 1996. Social interaction on the net: virtual community as participatory genre. Proceedings of the 30th Hawaii International Conference on Systems Sciences, Hawaii, January 1996. Ericsson and Simon, 1993 Ericsson K.A. Simon H. , Protocol Analysis: Verbal Reports as Data 2nd ed. 1984 1993 MIT Press , Cambridge, MA Goffman, 1959, 1969 Goffman E. , The Presentation of Self in Everyday Life 1959 1969 Penguin Press , London Goffman, 1981 Goffman E. , Forms of Talk 1981 Basil Blackwood , Oxford, UK Graham-Cumming, 1997 Graham-Cumming, J., 1997. Hits and Miss-es: A Year of Watching the Web, Proceedings of the Sixth International WWW conference, Santa Clara, CA, USA, April 1997. GVU, 1998 GVU (1998). 10th survey, GVU WWW User Surveys, www.gvu.gatech.edu/gvu/user_surveys/survey-1998-10 as on 2/2/2000. Hammond and Allinson, 1989 Hammond, N., Allinson, L., 1989. Extending hypertext for learning: an investigation of access and guidance tools in people and computers V.In: Sutcliffe, A., MaCauley, L. (Ed.). Proceedings of the 5th BCS Human–Computer Interaction SIG, Nottingham, September 1989, pp. 273–304. Levinson, 1988 Levinson S.C. , Putting linguistics on a proper footing: explorations in Goffman's concepts of participation Drew P. Wootton A. Erving Goffman: Exploring the Interaction Order 1988 Polity Press , Oxford, UK 161 227 OpenURL Placeholder Text WorldCat Light, 1999 Light A. , Vermersch's explicitation interviewing technique used in analysing human–computer interaction , CSRP 513 ( 1999 ) 1999 OpenURL Placeholder Text WorldCat Light and Rogers, 1999 Light A. Rogers Y. , Conversation as publishing: the role of news forums on the web , Journal of Computer-Mediated Communication on Persistent Conversation 4 ( 4 ) 1999 (special issue) www.ascusc.org/jcmc/vol4/issue4/ OpenURL Placeholder Text WorldCat McAleese, 1993 McAleese R. , Navigation and browsing McAleese R. Hypertext in Hypertext: Theory into Practice 1993 Intellect , Oxford, UK OpenURL Placeholder Text WorldCat Morse, 1998 Morse M. , Virtualities: Television, Media Art and Cyberculture 1998 Indiana University Press , Bloomington, IN (chap. 1) Nielsen, 1990 Nielsen J. , Hypertext and Hypermedia 1990 Academic Press , San Diego, CA Norman, 1988 Norman D.A. , The Psychology of Everyday Things 1988 Basic Books , New York Payne, 1990 Payne S.J. , Looking H–CI in the I in Interact 90 — Human–Computer Interaction 1990 Elsevier , Amsterdam Pemberton, 1996 Pemberton L. , Telltales and overhearers: participant roles in electronic mail communication Connolly J.H. Pemberton L. Linguistic Concepts and Methods in CSCW 1996 Springer , London 145 162 OpenURL Placeholder Text WorldCat Potter and Wetherell, 1987 Potter J. Wetherell M. , Discourse and Social Psychology: Beyond Attitudes and Behaviour 1987 Sage , London Reeves and Nass, 1996 Reeves B. Nass C. , The Media Equation: How People treat Computers, Television and New Media like Real People and Places 1996 Cambridge University Press , Cambridge, UK Sproull and Kiesler, 1992 Sproull L. Kiesler S. , Connections: New Ways of Working in the Networked Organisation 1992 MIT , Cambridge, MA Vermersch, 1994 Vermersch P. , L'Entretien d'Explicitation 1994 ESF , Paris Vermersch, 1999 Vermersch P. , Journal of Consciousness Studies 6 ( 2-3 ) 1999 ) 17 42 Winograd, 1987 Winograd T. , A language/action perspective on the design of cooperative work , Human–Computer Interaction 3 ( 1 ) 1987 ) 3 30 Google Scholar Crossref Search ADS WorldCat Wright, 1984 Wright P. , Informed Design for Forms Easterby R. Zwaga H. Information Design: the Design and Evaluation of Signs and Printed Material 1984 Wiley , Chichester, UK 545 577 OpenURL Placeholder Text WorldCat © 2001 Elsevier Science B.V. All rights reserved.
TI - Beyond the interface: users' perceptions of interaction and audience on websites
JO - Interacting with Computers
DO - 10.1016/S0953-5438(00)00044-8
DA - 2001-02-01
UR - https://www.deepdyve.com/lp/oxford-university-press/beyond-the-interface-users-perceptions-of-interaction-and-audience-on-gribdm0GgM
SP - 325
EP - 351
VL - 13
IS - 3
DP - DeepDyve
ER -